基于FRWT的模拟电路早期故障诊断

针对模拟电路早期故障诊断的难题,基于分数阶小波转换(fractional wavelet trarsform,FRWT)并结合隐马尔科夫模型(hidden Markov model,HMM),提出了一种模拟电路故障特征分析的新方法。首先将无故障状态和各故障状态下模拟待测试电路(circuit under test,CUT)的响应序列进行分数阶小波分解得到子带响应序列,然后从子带响应序列提取出故障特征向量并构成观测序列训练出HMM,最后利用训练好的HMM对未知状态电路进行诊断。实验结果表明,该方法能有效提取模拟电路的故障特征,完成模拟电路早期故障检测和故障定位。

Incipient fault diagnosis of analog circuits based on FRWT

Aiming at the problem of incipient fault diagnosis in analog circuits,based on fractional wavelet transform(FRWT) combined with hidden Markov model(HMM),a new approach is proposed to analyze the fault signatures of analog circuits.Firstly,the response sequences of the analog circuit under test(CUT) in fault-free state and faulty states are decomposed using fractional wavelet to obtain the response sequences in subbands.Then,the fault signature vectors extracted from the response sequences in subbands are used to form the observation sequences to train the HMMs of the CUT.Finally,the unknown states of the CUT are diagnosed using the well-trained HMMs.Experiment results show that the method proposed in this paper can extract the fault signatures of analog circuits effectively and achieve detection and location of incipient faults in analog circuits.